You Cannot Control What You Cannot See — The Origin Story
Every failed cloud cost initiative I've seen in my FinOps years had the same root cause. It wasn't the tooling. It wasn't the engineers. It wasn't even the budget.
It was simpler than that: no one had cloud cost visibility into the same picture.
I learned this the hard way—through three projects that had nothing in common on the surface. Different clouds, different industries, different teams. Let me show you what I mean. Three projects, three industries, one invisible thread. Yet every single post-mortem ended with the same uncomfortable sentence: we didn't see it.
Example 1: the "savings plan" that saved nothing
The organisation did what every "optimisation" playbook recommends: commit, get the discount, lock in the win.
On paper, it looked perfect.
In reality, the commitment was mapped to an architecture that was already changing. Workloads moved. Utilisation shifted. The discount applied to the wrong shape of consumption. The result was a monthly conversation that sounded like this:
Finance: "We bought savings. Where are the savings?"
Engineering: "We didn't change anything—why did cost go up?"
Procurement: "The contract says we should be cheaper."
Everyone had a dashboard. Nobody had visibility into their cloud spend.
Visibility is not a chart—it's shared, explainable truth: what are we consuming, why are we consuming it, and who can change it?
Example 2: the cost allocation model that created politics
Another organisation decided the real problem was chargeback. "If teams pay for what they use, behaviour will change."
So they invested months into allocation rules, tagging conventions, and internal billing.
When allocation is wrong, every number becomes a negotiation. Every spike becomes politics. Teams stop asking "what happened?" and start asking "who is going to be blamed?"
That is the moment you lose.
If the model is not trusted, the organisation cannot steer. Not because people are stubborn, but because they are rational: no one will act on data they don't believe.
And yes: chargeback only works after showback. If you can't see where money is going, all you can do is negotiate the bill.
Example 3: the "FinOps dashboard" nobody used
The third one was the most expensive. Not in dollars—though that too—but in opportunity cost.
A beautifully designed dashboard. Multiple dimensions. Trend lines. Forecasts. Executive summaries. It was everything a stakeholder could ask for.
And then the simplest question in the room killed it:
"Where does this number come from?"
No one could answer with confidence.
So the dashboard became decoration.
A dashboard has to tell a story. Not "what you can see", but what you should conclude from what you see.
Visibility is capability zero
This is the origin story behind my signature line: you cannot control what you cannot see.
In FinOps, people often want to start at the end: optimisation. Commitments. Rightsizing. Discounts. Negotiation.
But steering requires a sequence. The order matters—and the FinOps Framework is explicit about it:
Inform — build shared, explainable truth (not dashboards)
Optimize — turn that truth into better architecture and decisions
Operate — make it stick through cadence, ownership, and governance
Most organisations attempt step three with step one half-built.
They try to govern what they cannot explain.
They try to optimise what they cannot attribute.
They try to negotiate what they cannot measure.
And then they blame the tool or the people doing their best.
Seeing is not reporting
Reporting tells you what happened.
Visibility tells you:
what happened
why it happened
who can change it
how to validate the change worked
It is not a dashboard project. It is an operating model.
It also requires a decision: one source of truth. Not perfect. Not immutable. But agreed. In practice: one data warehouse, one monthly cost review, one owner — imperfect but agreed.
Because the moment every team brings their own numbers, you have already lost the steering wheel.
The uncomfortable takeaway
If you want a single practical test, here it is:
The next time someone points at cloud spend and asks "what happened?", measure how long it takes to answer with confidence—and whether three people give the same answer.
If the answers diverge, the problem is not optimisation. It is visibility.
And until you fix that, every "cost initiative" becomes expensive tuition for the same class.
